Contractors' risk assessment system

Risk in the construction industry greatly affects the reputation and the productivity rate of the industry. Construction surety businesses experience significant risk and uncertainties in their bonds underwriting operation. The qualification process in the construction bond business is very complicated, labor-intensive, and time-consuming. The construction bonds underwriting processes also contain subjective judgments and otherwise non-quantifiable elements that require the human involvement to reach decisions. In doing these analyses, the surety company's underwriters determine the level of risk that contractors represents according to the evaluation analyses for bonds awarded for potential projects. Thus, underwriting evaluations are bound to be very difficult, ambiguous, and have potentially great losses in business. On the other hand, the surety industry experienced significant impact on their profitability due to the financial losses incurred in the recent years because of the poor judgements made by surety underwriters and other risk factors. Therefore, this research investigated the construction bonds underwriting knowledge in evaluating contractors for potential projects. Then, the research proposed a new structured methodology or a model for the construction bonding evaluation. The new structured evaluation methodology has been implemented by the Contractors Risk Assessment System (CRAS) in which the system captured the knowledge of the experienced underwriters. The specific focus of CRAS is to judge or classify the risks of the construction bonds' applicants (contractors) based on the evaluation of a set of decision factors such as the four C's (character, capital, capacity, and continuity) and other relevant measures of risk. CRAS consists of four independent pattern classification modules for the 4 C's and an integrated pattern classification model to accomplish the focus of this research. The functions of the CRAS independent modules were to capture the experiences of assessing the risk associated with each decision factor of the qualification process separately and to adapt new experiences. The functions of the CRAS integrated model were to capture the experiences of assessing the overall risk of bonds applicants (contractors) by looking at the 4 C's decision factors together and to adapt new experiences. Backpropagation neural networks have been used as the pattern classification tool. A justification for neural network and backpropagation algorithm selection is presented. CRAS modules and the integrated model have been validated in which very successful results were achieved. The use of CRAS prototype would help underwriters to assess the significant risks of bonds' seekers to improve the quality of the pre-qualification analysis. Moreover, CRAS has overcome the underwriting highly unstructured environment, the process qualitative parameters in which it is difficult to evaluate or obtain precise data in a standard format, and the subjectivity nature of this problem. As a result, this research demonstrated that CRAS's new model has handled this complex and extremely subjective evaluation more efficiently to reach more reliable decisions.